Skip to content

/adr-plan

Analyze a task and produce an Architecture Decision Record with implementation steps.

shell
$ npx -y skills add NikiforovAll/claude-code-rules --skill adr-plan --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/adr-plan
How auto-invocation works

Context preview

The summary Claude sees to decide when to auto-load this skill.

Analyze a task and produce an Architecture Decision Record with implementation steps.

SKILL.md

adr-plan.SKILL.md
name: adr-plan
description: Analyze a task and produce an Architecture Decision Record with implementation steps.
allowed-tools: Agent, AskUserQuestion, Bash(npx adr *), Bash(git diff *), Bash(git log *), Bash(git status *), Bash(ls *), Bash(cat *), Read, Grep, Glob

ADR Plan: Task Analysis → Architecture Decision Record

Analyze a task, explore the codebase, and produce an ADR with concrete implementation steps.

Phase 1: Analyze

1. Read the task description, active plan, or task list 2. Explore affected areas of the codebase — do it concurrently for independent modules 3. Map blast radius — search for consumers of functions/types/routes being changed 4. Identify alternatives worth considering (at least 2)

Do this silently.

Phase 2: Detect ADR Setup

Check if the project has an ADR directory:

ls docs/adr/ || ls adr/ || ls doc/adr/
  • **Found** → use existing directory, detect next number from existing files
  • **Not found** → run `npx adr init en`, then proceed

Phase 3: Produce ADR Content

Write the ADR using `npx adr new "<title>"`, then edit the generated file with the following structure:

# ADR-NNNN: [Title]

## Status
Proposed

## Context
[What problem are we solving? What constraints exist?]

## Decision
[What we chose and why]

## Alternatives Considered
| Option | Pros | Cons |
|--------|------|------|
| ... | ... | ... |

## Implementation Steps

### Step 1: [Description]
- **Files:** [files to create/modify]
- **Depends on:** [previous step or "none"]
- **Done when:** [concrete acceptance criteria]

### Step 2: [Description]
...

## Consequences
- [Positive and negative outcomes, tradeoffs accepted]

Guidelines

  • **Steps are ordered by dependency** — each step lists what it depends on
  • **Steps are parallelizable when independent** — note which steps can run concurrently
  • **Each step has concrete "done when" criteria** — no vague outcomes
  • **Alternatives table is honest** — include the option you chose and why others lost
  • Keep it short. 1-2 pages max. No padding.

Phase 4: Present to User

Show the ADR content and the file path. The user may:

  • **Approve** → ADR stays as-is
  • **Adjust** → edit and re-present
  • **Cancel** → delete the file

Important

  • The ADR is a planning artifact, not documentation for posterity
  • Steps should map naturally to work units (a team member could own one or more steps)
  • If the task is too simple for an ADR (single file, obvious fix), say so and skip
  • Do not write implementation code — this is planning only
Read more
Read it on GitHub ↗
Ships withclaude-code-rules

A collection of Claude Code recommendations and practices. Learn practical techniques to enhance your AI-assisted development workflow with Claude Code.

Get the whole plugin, auto-invoked
Stats
136
Stars
0
Views
19
Forks
Active
Maintenance
HTML
Language
Apache-2.0
License
5d ago
Last commit
1y ago
Created

Repo: NikiforovAll/claude-code-rules

Other skills on claude-code-rules.